Fixed Rate in Paris, Adjustable Rate in Washington

A household rejects a fixed-rate loan as too expensive and opts for a variable rate instead. It immediately saves 1.2 percentage points—provided interest rates do not rise.

In many respects, that is precisely what the US government is doing.

The numbers add up

As of August 4, 2026, borrowing for one year costs approximately 4.00% in the United States. Borrowing for thirty years costs around 5.20%. That is a difference of 120 basis points.

Across hundreds of billions of dollars, the immediate savings amount to billions in interest. The calculation is tempting in a budget where debt service already absorbs 18.5% of federal revenue.

No debt manager would turn down such a short-term saving. Nor did the Treasury.

Documents released for the third quarter of 2026 project $739 billion in privately held net marketable borrowing. Alongside that figure, the issuance tables show approximately $375 billion in coupon-bearing securities and $409 billion in short-term Treasury bills. These amounts cannot simply be added together, because they do not cover the same accounting scope as quarterly net borrowing. Source

The message nevertheless remains clear: a substantial share of US financing relies on securities whose cost will soon be repriced by the market.

Repeated for long enough, this strategy transforms the structure of the risk.

As of July 31, 2026, Treasury bills with maturities of less than one year account for approximately $7 trillion, or nearly 22.2% of marketable debt held by the public. Source

Creditors have sounded the warning

Since 2020, the Treasury Borrowing Advisory Committee has recommended that the Treasury keep bills within a range of 15% to 20% of marketable debt. The committee brings together professionals from major banks, investment funds and bond markets. It advises the Treasury quarterly, and its reasoning has remained consistent: bills often cost less in the short term, but they expose the budget to a rapid rise in interest rates. Such an increase would feed through progressively with each refinancing, rather than being held at bay by locking in medium- or long-term funding.

The proposed solution is not necessarily to finance everything for thirty years. Rather, it is to increase the share of intermediate maturities—two or three years, for example—in order to reduce dependence on maturities of just a few weeks or months.

A three-month bill matures and must be refinanced quickly. A three-year note provides greater visibility: its coupon is fixed for the next three years.

That protection, however, comes at a price. Under normal conditions, yields on two- or three-year securities may exceed bill yields by several dozen basis points. That additional yield pays for a form of protection against refinancing risk. It is obviously not free insurance, since it incorporates interest-rate expectations and the term premium.

The US Treasury is therefore accepting an immediate saving in return for greater exposure to future interest-rate increases.

The average conceals the risk

One might object that US debt is not particularly short-dated and that its weighted average maturity is approximately six years. True. But an average combines maturities with very different risk profiles.

A debt stock composed partly of bills and partly of ten-, twenty- or thirty-year securities can display a high average maturity while remaining heavily exposed to short-term refinancing. The risk therefore depends not only on the average, but necessarily on the precise distribution of maturities.

That said, the United States does not have half of its debt in three-month paper. Although relatively high, the share of bills remains around one-fifth of marketable debt—not one-half.

The issue, therefore, is not that US debt is entirely “floating-rate,” but that its short-term share is large enough for higher interest rates to feed rapidly into the federal budget.

Paris made a different choice

France has favoured longer maturities. The average maturity of French government marketable debt is around eight and a half years—more precisely, eight years and 180 days as of June 30, 2026. Source

This does not mean that exactly one-eighth of the debt returns to the market each year: an average alone can never describe the actual maturity schedule. Nevertheless, this structure slows the transmission of an interest-rate shock to the public finances.

For illustration, if we assume that the debt is refinanced gradually and that interest rates rise uniformly by 100 basis points, the additional interest burden could reach approximately €3.1 billion in the first year, then accumulate to around €32.1 billion by the end of a decade. These figures are the result of a simulation, not a forecast, and assume, in particular, a stable debt stock and an even distribution of refinancing.

Yet this protection has a price: France’s projected debt-interest bill for 2026 is €59.3 billion—more than the budgets of many core government functions.

France therefore knows the price of its insurance.

America’s risk

Washington is currently enjoying the short-term funding discount, paying less interest than it would if it locked in all its borrowing at long maturities.

In return, the budget becomes more sensitive to the path of interest rates. A sustained increase would not hit the entire debt stock on the same day, but would spread rapidly with each rollover of Treasury bills.

A borrower who does not hedge is not necessarily wrong. If interest rates fall, or remain stable for a prolonged period, preferring short-term funding will have proved to be the right decision.

Yet this strategy ceases to be a mere debt-management choice once exposure to short-term rates reaches several trillion dollars. It becomes a macroeconomic decision—one that must be politically owned—and deserves to be explicitly set out in budget documents.

The problem is not that the United States borrows at short maturities. It is that US policymakers do not clearly disclose the price of the risk they are taking.